The Swarm Age: Drones, Robots and the New Calculus of Military Power

Somewhere along the contact line in eastern Ukraine, a pilot wearing goggles sits in a basement, thumbs on a controller that could have come from a gaming shop. The first-person-view quadcopter he steers costs around $400 to build. Its target, lumbering across a treeline gap, is a Russian main battle tank worth several million dollars. The drone dives, detonates and the thermal footage circulates online within hours. Ukraine now produces these weapons at a rate exceeding 200,000 per month, up tenfold from January 2024, with more than 160 manufacturers feeding the ecosystem.

Three thousand kilometres to the south, the same arithmetic is playing out at strategic scale. After coordinated US-Israeli strikes on Iran on 28 February 2026, Tehran retaliated with sustained waves of Shahed drones and ballistic missiles aimed at Israel, US bases and every Gulf state seen as allied with Washington. Within the first week Iran had launched more than 1,000 drones; by late March the running total at Gulf targets alone had passed 3,000 drones and roughly 1,000 ballistic missiles. Each Shahed-136 derivative costs Iran somewhere between $20,000 and $70,000; each Patriot PAC-3 interceptor sent up to meet it costs around $4 million, with THAAD rounds closer to $12 million. By the end of March, the Jewish Institute for National Security of America estimated that the UAE and Kuwait had burned through some 75 per cent of their Patriot stocks, with Bahrain at 87 per cent. The Pentagon told Congress the first six days of the war alone had cost the United States $11.3 billion. By April, Saudi Arabia and the UAE were in active talks to buy a $2,500 Ukrainian-designed interceptor drone to do the work their million-dollar missiles could no longer afford.

The same maths has been grinding away in the Red Sea, on a longer timeline. US Navy destroyers have spent more than 15 months intercepting waves of cheap drones and cruise missiles launched by Houthi forces. A single SM-2 interceptor costs roughly $2.1 million. Many of the targets it destroys are Iranian-designed Shahed-136 derivatives estimated at $20,000 to $50,000 each. By early 2025, the US Navy had fired more than 200 surface-to-air missiles and the Eisenhower carrier strike group alone had expended 770 weapons on a single deployment. Senior Pentagon officials have described the expenditure rate as unsustainable.

These three vignettes are not isolated curiosities. They are data points in a structural shift that spans the full spectrum of uncrewed systems, from $400 FPV quadcopters to multi-tonne collaborative combat aircraft designed to fly alongside fifth-generation fighters. The old mathematics of military power, in which the defender's technological edge reliably outweighed the attacker's numbers, no longer holds. The age of the swarm is arriving at every altitude.

Figure 1. National Capability Ratings Distribution for Drones, Swarming & Collaborative Robots

From Hobby Shop to Battlefield

The drone's journey from novelty to strategic weapon has been compressed into barely two decades, and it has proceeded along two distinct tracks that are now converging.

The first track was exquisite and expensive. The MQ-1 Predator and its successor, the MQ-9 Reaper, were operated almost exclusively by the United States and a handful of trusted allies. At roughly $30 million per aircraft (with ground station and sensors), the Reaper required dedicated satellite links, elaborate ground control and years of pilot training. It was, in every meaningful sense, a crewed platform with the cockpit removed. The Reaper remains the backbone of Western long-endurance ISR and strike; more than 300 are in service across the US, UK, France, Italy and others. But its operational concept, a single high-value asset loitering for 24 hours over permissive airspace, looks increasingly vulnerable in contested environments saturated with air defences and electronic warfare.

The second track was cheap and disposable. The inflection point came in autumn 2020, when Azerbaijan used Turkish Bayraktar TB2 drones and Israeli IAI Harop loitering munitions to dismantle Armenian armoured formations in the Second Nagorno-Karabakh War. Ukraine then changed the scale entirely. By 2024 it had produced roughly 2.2 million UAVs, of which more than 1.5 million were FPV combat drones. The 2025 target is 4 to 4.5 million units, with projected capacity of over 8 million FPV drones by 2026.

At the centre of the proliferation story sits Iran. The Shahed-136 has been supplied to Russia, to the Houthis and to Hezbollah. Russia has licensed production at the Alabuga special economic zone in Tatarstan, scaling output from 250 units per month in 2023 to an estimated 5,000 Geran-2 drones per month by late 2025. Iran has demonstrated that a sanctioned middle power can reshape global drone markets through attritable mass and strategic licensing.

Now a third track is emerging: collaborative combat aircraft (CCA) designed to operate as autonomous wingmen alongside crewed fighters. These are not small drones. The US Air Force has downselected Anduril's Fury (YFQ-44A) and General Atomics' Gambit (YFQ-42A) for its CCA programme, with both achieving first flights in 2025. Australia's MQ-28 Ghost Bat, Boeing's stealthy loyal wingman, has completed more than 150 test flights and fired a live air-to-air missile. These platforms sit at the high-value end of the spectrum, but they are designed to be cheaper than the crewed fighters they accompany and, critically, expendable if the mission demands it. The DSAC landscape now stretches from hobby-shop FPVs through industrial-scale Reapers to jet-powered autonomous wingmen, and the countries that master all three tiers will hold a decisive advantage.

Measuring Capability

For all the attention drones now command, no existing assessment framework captures the full picture. The IISS Military Balance catalogues platform inventories across 170 countries. Its coverage of MALE and HALE classes (Reaper, Heron, Wing Loong) is thorough, but it does not systematically track the tactical drones that dominate the battlefield, nor does it assess autonomous swarming capability or CCA programmes. SIPRI's Arms Transfers Database tracks UAV exports but applies a 20-kilogram minimum threshold, excluding FPV munitions entirely and classifying one-way attack drones as missiles rather than aircraft.

Specialist trackers fill niches. Drone Wars UK monitors the proliferation of armed MALE drones across 48 states. The New America Foundation built databases of US drone strikes in Pakistan, Yemen and Somalia. The Teal Group projects $258.7 billion in worldwide military UAS spending over the coming decade. Janes profiles more than 500 platforms. CSIS and RUSI have produced sharp operational analysis of Ukraine's drone war. Each illuminates part of the landscape while leaving significant terrain in shadow.

Where Nations Stand

The distribution of GINC DSAC ratings is heavily concentrated at the lower end of the scale, with 128 nations at FFF, reflecting the reality that most countries have no meaningful indigenous capacity in collaborative autonomous systems. Only three nations reach the Frontier tier (A): China, the United States and Ukraine, each arriving by a distinct pathway: full-stack industrial sovereignty, systems integration leadership and combat-driven iteration respectively.

Figure 1. GINC Ratings.

The Advanced band (BBB through B) contains 13 nations, but even here the gap between operational leaders such as Iran and Israel and industrial-base holders such as Germany and the Netherlands is substantial. This is among the most unevenly distributed capabilities in the GINC framework, and one where the correlation between broad critical-technology depth and realised operational capability is unusually weak. Iran, a sanctioned middle-income economy, rates BBB alongside Israel and South Korea. The United Kingdom, with the world's sixth largest defence budget, sits at B. Combat experience and industrial agility matter more here than GDP.

Two nations carry Upgrade outlooks. Australia's Ghost Bat programme, backed by A$1.4 billion in new investment and growing export interest from Germany, Japan and the United States, positions it to move from B to BB as the platform enters service in 2028. The UAE's investment in indigenous drone production and counter-UAS directed energy is similarly on an upward trajectory. Taiwan's Upgrade outlook at CCC reflects accelerating domestic drone programmes driven by the cross-strait threat.

The distribution of national capability ratings in Drones, Swarming and Collaborative Robots is heavily concentrated at the lower end of the GINC rating scale, with 128 nations at FFF, reflecting the reality that most countries have no meaningful indigenous capacity in collaborative autonomous systems. The sharp spike at FFF and the secondary cluster at E (27 nations) illustrate a capability that remains the preserve of a small number of states. Only three nations reach the Frontier (A): China, the United States and Ukraine, each arriving there by a distinct pathway, full-stack industrial sovereignty, systems integration leadership and combat-driven iteration respectively. The Advanced band (BBB through B) contains 13 nations, but even here the gap between operational leaders such as Iran and Israel and industrial-base holders such as Germany and the Netherlands is substantial. This is among the most unevenly distributed capabilities in the GINC framework, and one where the correlation between broad critical technology depth and realised operational capability is unusually weak.

Ten Countries, Four Mechanisms

Assessing national capability in Drones, Swarming and Collaborative Robots requires looking beyond headline platform counts or R&D spending. The decisive question is whether a nation can close the loop between designing a system, teaching it to coordinate with others, keeping it connected under pressure, commanding it intelligently, and learning from its performance in the field. The GINC framework decomposes DSAC capability into five sub-capacities that together describe this loop. A nation strong in only one or two will have visible programmes and impressive demonstrations, but will not achieve frontier status. The nations that reach the top of the rating scale are those that sustain meaningful capacity across all five.

Production is the industrial substrate: the capacity to design, manufacture and sustain unmanned aerial, ground, surface and subsurface platforms at scale. It covers component sovereignty (motors, batteries, flight controllers, semiconductors), airframe and propulsion manufacturing, and the resilience of supply chains against export controls or single-source dependencies. Production is where the visible asymmetries lie — China's dominance of commercial drone manufacturing, Iran's scaling of one-way attack munitions under sanctions, Ukraine's distributed workshop network — and it is the sub-capacity most closely tied to hard industrial policy.

Intelligence is the software and algorithmic layer that turns collections of platforms into collectives: distributed decision-making, consensus protocols, collision avoidance, dynamic task allocation, and the emergent behaviours that distinguish a swarm from a formation. It includes on-board artificial intelligence, edge computation, and the simulation environments used to train and validate behaviours before deployment. This is where the quality gap between genuine swarming and choreographed multi-platform operation becomes visible, and where the United States currently holds its clearest lead.

Connectivity is the nervous system: the mesh networking, low-latency datalinks, frequency-hopping and anti-jamming capacities that allow collaborative robots to maintain coherence under contested electromagnetic conditions. It encompasses resilience to GNSS denial, cyber intrusion and adversarial spoofing. Connectivity is increasingly the decisive factor separating laboratory swarms from operationally credible ones, and it is where several highly rated research nations have unexpected weaknesses relative to combat-tested peers.

Control is the human–machine teaming dimension: the doctrines, interfaces and command architectures through which operators task, supervise and intervene in autonomous collectives. It covers operator-to-swarm ratios, mission planning tools, trust calibration and the integration of unmanned systems into existing command structures across military, civil and dual-use domains. Control is where doctrinal maturity matters as much as technology, and where nations with strong industrial bases but limited operational experience tend to lag.

Deployment is the capacity to field collaborative robotic systems in real environments — combat, disaster response, infrastructure inspection, logistics — and to iterate rapidly on operational feedback. It includes test ranges, doctrinal experimentation units, and the feedback loops connecting frontline use to industrial and algorithmic refinement. Deployment is the sub-capacity where Ukraine has most dramatically outpaced its pre-war critical technology base, and where the framework most clearly rewards learning under pressure over peacetime preparation.

The table below scores the leading nations across each of these five sub-capacities on a 1–10 scale, with the overall DSAC score derived as an unweighted mean. The divergences between sub-capacity scores are often as revealing as the totals: a nation strong on production and deployment but weak on intelligence and connectivity tells a very different strategic story from one with the reverse profile.

United States. The Replicator programme, launched in 2023 to field thousands of autonomous systems, delivered hundreds before being restructured into a counter-UAS task force. The more consequential bet is CCA: 100 to 150 aircraft planned for Increment 1, with $804.4 million budgeted in FY2026. The US also operates the world's largest Reaper fleet and leads in high-end autonomy research. But its defence industrial base is not structured to produce cheap drones in Ukrainian volumes.

China. The Zhuhai 2024 airshow revealed the Jetank swarm mothership (25-metre wingspan, 16-tonne takeoff weight), the stealthy CH-7 UCAV and the long-endurance CH-9 (40-hour flight time, 11,500-kilometre range). Chinese firms control roughly 70 per cent of the global commercial drone market and production capacity is estimated at a significant multiple of American output. China scores strongly across all four mechanisms.

Ukraine. More than 160 manufacturers produce everything from $400 FPV quadcopters to $200,000 long-range strike platforms. On track for 4 to 4.5 million drones in 2025. Fly It and Make It scores are exceptional; Swarm It and Team It remain nascent, with most operations still single-operator engagements.

Russia. The Alabuga facility produces upwards of 5,000 Geran-2 drones per month. Russia also fields the ZALA Lancet and operates Orion MALE drones. Make It and Fly It are strong through sheer volume; swarming and teaming capabilities remain rudimentary.

Israel. A pioneer since IAI developed the Harop, Israel combines deep sophistication with combat experience. Iron Beam, operational since December 2025, intercepts at roughly $2 to $5 per shot versus $40,000 to $50,000 for Iron Dome. Elbit's SkyStriker and LANIUS add swarming capability. Strong on Team It and Fly It; Make It constrained by a small industrial base.

Turkey. Baykar's TB2 has been delivered to 34 countries, with $2.2 billion in export revenue in 2025 and an estimated 60 per cent of the global armed-drone export market. The jet-powered Kızılelma signals Turkey's move from MALE platforms toward combat-capable UCAVs.

Iran. Tehran's influence is disproportionate to its GDP. The Shahed-136 has armed Russia, the Houthis and Hezbollah. Innovation lies in attritable design philosophy rather than technological sophistication. Make It is high relative to the economy; Swarm It and Team It are limited; the Fly It portfolio is narrow, dominated by one-way attack platforms.

United Kingdom. StormShroud, an EW drone fielded by RAF 216 Squadron in 2025, and the £4 billion SDR commitment to autonomous systems mark progress. The UK operates Reaper (scheduled for replacement), has cancelled Project Mosquito but is pursuing CCA through international partnerships. Ambition outpaces fielded capability.

Australia. The MQ-28 Ghost Bat is the most advanced loyal wingman outside the US: 150+ test flights, first live missile firing in December 2025, RAAF service entry targeted for 2028. Export interest from Germany, Japan and the US is growing. Leads on Team It but limited on Fly It and Make It.

South Korea. Hanwha has committed 750 billion won ($524 million) to co-develop the Grey Eagle STOL with General Atomics. Korean Air leads the indigenous KUS-FS MALE drone. The chaebol manufacturing base gives strong Make It potential; combat-tested Fly It experience remains limited.

Trending Up, Trending Down

Several capabilities are ascendant. Attritable mass production sits at the top: the ability to manufacture expendable drones in the hundreds of thousands and treat them as munitions. Ukraine has proved the concept; China and Iran have industrialised it.

Collaborative combat aircraft represent the high-end growth area. The convergence of AI-enabled autonomy, low-observable airframes and open mission systems is producing a new class of platform that can fly alongside F-35s and Typhoons, carry weapons, extend sensor networks and absorb risk that would otherwise fall on crewed aircraft. The US, Australia, the UK and several European nations are investing heavily. CCA is to the 2030s what Reaper was to the 2000s: the platform that redefines what air forces look like.

Counter-UAS directed energy is moving from laboratory to fielding. Israel's Iron Beam represents the first generation. Electronic warfare systems are evolving from vehicle-mounted to trench-level. FPV and one-way attack munitions continue to proliferate, with fibre-optic guidance circumventing RF jamming. Maritime and subsea uncrewed systems, demonstrated by Ukraine's naval drones against the Black Sea Fleet, are an emerging domain.

On the other side of the ledger, exquisite single-platform procurement looks mismatched to a threat environment defined by volume and speed. Legacy MALE drones like the Reaper, while still operationally valuable for ISR in permissive airspace, face growing risk in contested environments and are being supplemented by cheaper, more survivable alternatives. Unarmoured surface vessels in contested littorals face existential risk. Static air defence without C-UAS layers is vulnerable to saturation. Single-role attack helicopters are being challenged by FPV drones at a fraction of the cost.

The common thread is a shift from platforms to portfolios, from unit capability to systemic capacity, from decade-long development cycles to months-long iteration.

Strategic Considerations

Cost-exchange ratios. When does the defender run out of interceptors before the attacker runs out of drones? Iran's April 2024 strike demonstrated a roughly 10:1 cost ratio in the attacker's favour. The Houthi campaign has pushed ratios beyond 200:1. Directed energy offers a partial answer, but fielding it at scale will take years.

Industrial mobilisation. Can Western defence primes, structured for low-rate production of high-specification platforms (whether Reapers or CCAs), match the cost curves of consumer electronics manufacturing? China's Shenzhen-based supply chain can produce drone components at scale, cost and speed that the West cannot yet replicate.

Autonomy and ethics. As autonomy stacks advance, the question of machine targeting authority moves from ethics seminars to operational doctrine. Current systems require a human operator for the final strike decision. Future swarming architectures and CCA concepts may compress that decision cycle beyond human reaction times.

Proliferation. Non-state actors now field capabilities that were, a decade ago, the preserve of advanced militaries. Export controls designed for Reaper-class platforms are poorly suited to dual-use components purchased on commercial marketplaces.

Counter-drone layering. Effective C-UAS requires electronic warfare, kinetic interceptors, directed energy and interceptor drones. Building and integrating these layers is the defining air defence challenge of the next decade.

After Ukraine and Iran: What Comes Next

The market is bifurcating. At the high end, the United States, United Kingdom, Australia and Israel are investing in CCA and human-machine teaming: autonomous wingmen that extend the reach of crewed fighters and represent the successor generation to platforms like the Reaper. At the volume end, China, Iran, Turkey and Ukraine are demonstrating that mass-attritable production reshapes battlefields and strategic calculations alike.

Counter-drone technology will be the growth sector through 2030. Iron Beam's debut signals a wave of directed-energy investment. Interceptor drones, cheap enough to trade one-for-one with attacking swarms, are in development across multiple countries.

The deepest shift may be in where value resides. Airframes, whether a $400 FPV or a multi-million dollar CCA, are becoming commoditised relative to the software that animates them. The autonomy stacks that enable collective behaviour, the sensor-fusion algorithms that allow target identification and the mission-planning systems that coordinate hundreds of platforms simultaneously are the new competitive moat.

Export controls will fragment further. Dual-use components are globally traded consumer goods. Doctrine will catch up, unevenly. Ukraine has embedded drone operators at every echelon from platoon to brigade. NATO allies are studying those lessons, but institutional inertia slows adoption.

DSAC is not a capability to bolt on to existing force structures. It is a test of whether defence establishments can think in volumes rather than platforms, iterate in months rather than decades, and accept that the most consequential weapon on tomorrow's battlefield may be either a quadcopter that costs less than a second-hand bicycle or a jet-powered wingman that fights alongside an F-35.